1. Filed of the Invention
The present invention relates to a fault tolerant communication control processor, which ensures high reliability in processing the transferred data as a whole system, even though a fault has occurred, in a time division multiplex data communication system.
In recent years, it has been necessary for relatively large amounts of data to be transferred with high speed and simultaneously with high reliability, because of the increase of computer networks. To meet the speed requirements the multiplex data communication system, such as an integrated service digital network (ISDN), is generally utilized. On the other hand, to meet the reliability requirement, a fault tolerant communication control processor that has a multiplexed and redundant circuit configuration, is generally utilized.
2. Description of the Related Art
In a computer network using a time division multiplex data communication system, in the prior art, a communication control processor (CCP) comprises a stand-by communication control unit (CCU) that is usually idle, and well as a normal communication control unit (CCU), which is currently used: namely, the duplexed and redundant configuration having normal CCU and stand-by CCU is adopted. In this configuration, even when the normal CCU has failed during transmission or the reception of data, the stand-by CCU may be selected and operated immediately, and therefore high reliability in the whole system may be maintained.
The above prior art will be explained more specifically. In this data communication system, it is assumed that a large amount of data are transferred and processed by the frame unit. Further, one frame is divided into a plurality of time slots of numbers n (n denotes natural number) and each time slot (TS) contains the data for one line, which is called a channel.
Therefore, in transmission side, the data for N channels are transmitted by dividing and multiplexing these data. In the reception side, these data are stored in the predetermined location by isolating the multiplexed time slots of every channel. If each frame data is transferred repeatedly with the required times, all the data can be transferred by the unit of each channel. Namely, all the data can be constructed by collecting the data isolated in the reception side of every channel.
A sequence of these operations are controlled by a normal CCU, which is assigned previously by the host computer. Further, a stand-by CCU is designed to have the same data as the normal CCP during the above operations, in order to be ready to start at any time. If some fault has occurred in at least one of the channels controlled by the normal CCU, all the channels are immediately switched, so that the controller of the channels may be transferred from the normal CCU to the stand-by CCU. Consequently, high reliability in the whole system including the normal CCU and stand-by CCU is ensured against accidental error.
In the prior art, however, there are the following problems.
First, even when the fault has occurred in only one channel, all channels, including the other normal channels, must also be switched. During the time for switching these channels, communication with the normal channels are stopped, which means that the faulty channel has a significant influence on the normal channels.
Second, when the normal CCU is normally operating, the stand-by CCU is idle. Therefore, the potential of a stand-by CCU in processing transferred data cannot be fully utilized, which leads to the deterioration of the efficiency of the data processing.
Third, since the stand-by CCU is not usually used, it is difficult to detect the fault existing in itself. Therefore, the above fault is not likely to be found until all channels are changed over from the normal CCU to the stand-by CCU, which leads to the deterioration of the reliability of the whole system.